Semi-supervised Segmentation Based on Error-Correcting Supervision

  • Pixel-level classification is an essential part of computer vision. For learning from labeled data, many powerful deep learning models have been developed recently. In this work, we augment such supervised segmentation models by allowing them to learn from unlabeled data. Our semi-supervised approach, termed Error-Correcting Supervision, leverages a collaborative strategy. Apart from the supervised training on the labeled data, the segmentation network is judged by an additional network. The secondary correction network learns on the labeled data to optimally spot correct predictions, as well as to amend incorrect ones. As auxiliary regularization term, the corrector directly influences the supervised training of the segmentation network. On unlabeled data, the output of the correction network is essential to create a proxy for the unknown truth. The corrector’s output is combined with the segmentation network’s prediction to form the new target. We propose a loss function that incorporates both the pseudo-labels as well as the predictive certainty of the correction network. Our approach can easily be added to supervised segmentation models. We show consistent improvements over a supervised baseline on experiments on both the Pascal VOC 2012 and the Cityscapes datasets with varying amounts of labeled data.

Export metadata

Additional Services

Share in Twitter Search Google Scholar Statistics
Author:Robert Mendel, Luis Antonio De Souza Jr.ORCiD, David Rauber, João Paulo PapaORCiD, Christoph PalmORCiDGND
Parent Title (English):Computer vision - ECCV 2020: 16th European conference, Glasgow, UK, August 23-28, 2020, Proceedings, Part XXIX
Place of publication:Cham
Document Type:conference proceeding (article)
Year of first Publication:2020
Release Date:2020/10/23
GND Keyword:Semi-Supervised Learning; Machine Learning
First Page:141
Last Page:157
Konferenzangabe:European Conference on Computer Vision, 16th, 2020
Andere Schriftenreihe:Lecture Notes in Computer Science; 12374
Institutes:Fakultät Informatik und Mathematik
Regensburg Center of Health Sciences and Technology - RCHST
Fakultät Informatik und Mathematik / Regensburg Medical Image Computing (ReMIC)
research focus:Digitalisierung
Licence (German):Keine Lizenz - Es gilt das deutsche Urheberrecht: § 53 UrhG